Selective Inhibition of Heterotrimeric GsSignaling

The blockade of heptahelical receptor coupling to heterotrimeric G proteins by the expression of peptides derived from G protein Gα subunits represents a novel means of simultaneously inhibiting signals arising from multiple receptors that share a common G protein pool. Here we examined the mechanism of action and functional consequences of expression of an 83-amino acid polypeptide derived from the carboxyl terminus of Gαs(GsCT). In membranes prepared from GsCT-expressing cells, the peptide blocked high affinity agonist binding to β2adrenergic receptors (AR) and inhibited β2AR-induced [35S]GTPγS loading of Gαs. GsCT expression inhibited β2AR- and dopamine D1Areceptor-mediated cAMP production, without affecting the cellular response to cholera toxin or forskolin, indicating that the peptide inhibited receptor-Gs coupling without impairing G protein or adenylyl cyclase function. [35S]GTPγS loading of Gαq/11 by α1BARs and Gαi by α2AARs and Gq/11- or Gi-mediated phosphatidylinositol hydrolysis was unaffected, indicating that the inhibitory effects of GsCT were selective for Gs. We next employed the GsCT construct to examine the complex role of Gs in regulation of the ERK mitogen-activated protein kinase cascade, where activation of the cAMP-dependent protein kinase (PKA) pathway reportedly produces both stimulatory and inhibitory effects on heptahelical receptor-mediated ERK activation. For the β2AR in HEK-293 cells, where PKA activity is required for ERK activation, expression of GsCT caused a net inhibition of ERK activation. In contrast, α2AAR-mediated ERK activation in COS-7 cells was enhanced by GsCT expression, consistent with the relief of a downstream inhibitory effect of PKA. ERK activation by the Gq/11-coupled α1BAR was unaffected by GsCT. These findings suggest that peptide G protein inhibitors can provide insights into the complex interplay between G protein pools in cellular regulation.

Heptahelical, or G protein-coupled, receptors represent the single most diverse class of cell surface receptors, both evolutionarily and within the human genome. The basic unit of G protein-coupled receptor signaling is composed of three parts as follows: a heptahelical receptor, a heterotrimeric G protein, 1 and an effector, such as a G protein-regulated enzyme or ion channel. The binding of an extracellular agonist ligand to the receptor changes its conformation so as to permit productive coupling with the G protein, thereby catalyzing the exchange of GTP for GDP on the G␣ subunit, and dissociation of G␣-GTP from G␤␥ subunits. Regulation of effectors is achieved through their interaction with free GTP-bound G␣ or G␤␥ subunits. Based upon data from crystallographic, biochemical, and mutagenesis studies, physical coupling of receptor and G protein is thought to involve primarily the second and third intracellular domains of the receptor, which make physical contact with the carboxyl terminus of the G␣ subunit (1)(2)(3)(4)(5)(6). In particular, the last ϳ50 amino acids of the G␣ subunit are important for discriminating between different receptor subtypes and between different functional states of the receptor (3, 4, 6 -9).
Pharmacologic agents that act as agonists or antagonists of heptahelical receptors represent the most common type of drug in clinical use today. Irrespective of chemical composition, these agents share a common mechanism of action in that they act extracellularly either to mimic, or to preclude, agonist binding at its receptor. By interacting with the molecular determinants of ligand binding in the extracellular or transmembrane domains of the receptor, often remarkable receptor subtypespecific agonist or antagonist effects can be obtained.
An alternative approach to antagonism of heptahelical receptor signaling is to target the receptor-G protein interface with agents that block coupling between the receptor and G protein intracellularly. Such an approach differs fundamentally from classical heptahelical receptor pharmacology in that the blockade of receptor-G protein coupling might be expected to produce G protein-specific, rather than receptor-specific, antagonism. Several successful applications of this strategy, using polypeptides derived from the putative contact surfaces on the receptor, or the G protein G␣ subunit, have been reported. For example, cellular expression of peptides derived from the third intracellular domains of the G q/11 -coupled ␣ 1B adrenergic receptor (AR) and M 1 muscarinic acetylcholine receptor, the G i -coupled ␣ 2A AR and M 2 acetylcholine receptor, and the G scoupled D 1A dopamine receptor have been shown to inhibit G q/11 -, G i -, and G s -coupled receptor signaling, respectively (10,11).
Analogous strategies have been applied using modified G␣ subunits or G␣ subunit-derived peptides. Cellular expression of a mutant G␣ s containing three point mutations that impair its function strongly inhibits G s -dependent stimulation of adenylyl cyclase in cultured cells (12). Modified xanthine nucleotidebinding mutants of G␣ o (13)(14) and G␣ 16 (15) inhibit signaling by G i -coupled receptors when expressed in COS-7 cells, whereas xanthine nucleotide-binding mutants of G␣ 11 and G␣ 16 (15) inhibit G q -coupled receptor signaling. Smaller peptides, derived from the carboxyl terminus of G␣ subunits, have been shown to produce similar inhibitory effects in membrane preparations and in intact cells (16 -20). Cellular expression of a minigene encoding the last 55 amino acids of G␣ q inhibits G q/11 -coupled receptor signaling (18). Minigene plasmids encoding oligopeptides representing the carboxyl termini of G␣ i , G␣ q , G␣ 12 , and G␣ 13 have recently been employed to determine the contribution of different G protein pools to signaling by M 2 muscarinic and thrombin receptors (19,20).
To determine whether an expressible peptide could be identified that interrupts signaling at the receptor-G s interface, we have prepared a series of minigene constructs encoding varying length polypeptides derived from the carboxyl terminus of G␣ s . In this paper, we characterize the mechanism of action and consequences of expression of an 83-amino acid G␣ s carboxylterminal polypeptide (GsCT). We find that the GsCT peptide selectively inhibits receptor-G s coupling in isolated plasma membranes and second messenger production in intact cells, without affecting G q/11 or G i signaling. When employed to examine the role of G s in regulation of the ERK MAP kinase cascade, we find that GsCT expression reveals both stimulatory and inhibitory effects of G s in response to activation of G s -, G i -, and G q/11 -coupled adrenergic receptors. These data indicate that expression of peptides derived from the carboxyl terminus of G␣ s can induce G protein-specific blockade of G s -coupled receptor signaling. By selectively blocking a single G protein pool, this approach can potentially provide insights into the contribution of different G protein pools to complex signaling processes.

EXPERIMENTAL PROCEDURES
Materials-HEK-293 and COS-7 cells were from the American Type Culture Collection. Tissue culture media, fetal bovine serum (FBS), geneticin (G418), and penicillin/streptomycin were from Invitrogen. FuGENE 6 was from Roche Molecular Biochemicals. 3Ј,5Ј-[ 3 H]cAMP was from Amersham Biosciences. myo-[ 3 H]Inositol and [ 35 S]GTP␥S were from PerkinElmer Life Sciences. Monoclonal anti-G␣ s , anti-G␣ i1/2 , and anti-G␣ q/11 IgG were from Calbiochem and PerkinElmer Life Sciences. Polyclonal anti-FLAG and anti-HA were from Santa Cruz Biotechnology. Anti-FLAG M2 affinity-agarose was from Sigma. Polyclonal anti-ERK1/2 and anti-phospho-ERK1/2 IgG were from Cell Signaling Technology. Horseradish peroxidase-conjugated donkey anti-mouse IgG was from Amersham Biosciences. Cholera toxin, H89, forskolin, isoproterenol, and 6-chloro-PB hydrobromide were from Sigma. Bordetella pertussis toxin was from List Biological. The cDNAs encoding the hamster ␣ 1B AR and the human ␣ 2A AR were provided by R. J. Lefkowitz. The cDNA encoding the human D 1A dopamine receptor was from M. G. Caron. The cDNA encoding the bovine G␣ s subunit was provided by A. G. Gilman.
Construction of Minigenes Encoding the G␣ s Carboxyl Terminus-The construction of the G␣ s peptide minigenes is depicted schematically in Fig. 1A. The PCR was employed to amplify cDNA encoding amino acids 286 -395, 313-395, or 337-395 of bovine G␣ s and the translation stop codon. Restriction sites at the 5Ј and 3Ј ends of the G␣ s -derived sequence were incorporated into the oligonucleotide primers used for DNA amplification. For the construct encoding a G s -derived peptide capable of post-translational prenylation, the DNA sequence TGCGTC-CTCTCTT, encoding the peptide sequence CVLS, was incorporated into the 3Ј end of the cDNA sequence prior to the stop codon. For the hemagglutinin (HA) epitope-tagged constructs, the G␣ s carboxyl-terminal sequences were subcloned as EcoRI to SalI fragments into a modified pcDNA3.1 that contained a Kozak sequence and an amino-terminal HA epitope upstream of the cloning sites. For the glutathione S-transferase (GST)-tagged constructs, G␣ s carboxyl-terminal sequences were subcloned into the pEBG vector, which encodes an aminoterminal GST epitope. The minigene plasmid encoding the third intracellular domain of the human D 1A dopamine receptor (D 1A 3i) in pRK5 was prepared as described previously (10).
Cell Culture and Transfection-COS-7 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% FBS and 50 g/ml penicillin/streptomycin. HEK-293 cells were maintained in minimum essential medium supplemented with 10% fetal bovine serum and 50 g/ml penicillin/streptomycin. Transient transfection of 40 -50% confluent cultures of COS-7 or HEK-293 cells in 100-mm dishes was performed using a ratio of 3 l of FuGENE 6 per g of plasmid DNA, according to the manufacturer's directions. Empty pcDNA3.1 vector DNA was added to each transfection as needed to keep the mass of DNA constant. A stable HEK-293 cell line expressing the GsCT minigene, G␣ s -(313-395), was prepared by calcium phosphate transfection using 5 g/ml G418 for selection, as described previously (21). Minigene expression following transient or stable transfection was detected by protein immunoblotting using antisera directed against the G␣ s carboxyl terminus. All assays on transiently transfected cells were performed after 48 -72 h. Prior to assay, transfected cells were split into multiwell plates, as appropriate, and incubated overnight in growth medium supplemented with 0.5% FBS and 10 mM HEPES, pH 7.4.
Competition and Saturation Binding Assays-Plasma membrane preparations for use in binding assays, [ 35 S]GTP␥S loading of G␣ subunits, and immunoblotting were prepared by differential centrifugation. Monolayers of appropriately transfected COS-7 or HEK-293 cells were scraped into 4°C lysis buffer (10 mM Tris-HCl, pH 7.4, 5 mM EDTA) and subjected to Dounce homogenization. Membranes were isolated by sequential centrifugation at 300 ϫ g for 3 min to remove cell nuclei and unbroken cells, and 40,000 ϫ g for 30 min to collect plasma membranes. The supernatant from the second centrifugation represented the cytosolic fraction. For ␤ 2 AR competition binding analyses, membranes were resuspended in binding buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM EDTA) at a concentration of 0.5 mg of protein/ ml. Membrane aliquots (20 g of protein) were incubated with [ 125 I]cyanopindolol for 30 min at 37°C in the presence of varying concentrations of isoproterenol (0 -10 Ϫ5 M) and then filtered over Whatman GF-C filters and washed to separate unbound ligand. Nonspecific binding was determined in the presence of 25 M alprenolol. To confirm that assays of GsCT effects were performed under conditions of equal receptor expression, the level of ␣ 1B AR, ␣ 2A AR, ␤ 2 ARonco, and D 1A dopamine receptor expression in HEK-293 and COS-7 cells was determined by saturation binding analysis, as described previously (11).
[ 35 S]GTP␥S Loading of G␣ Subunits-Assays of [ 35 S]GTP␥S loading of endogenous G␣ s , G␣ q/11 , and G␣ i1/2 subunits were performed on cell membranes prepared from HEK-293 cells. Membrane pellets were resuspended in TME buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.4) containing 2 mM EDTA and 4.8 mM MgCl 2 for G s loading assays or 100 M EDTA and 120 M MgCl 2 for G q/11 and G i loading assays. [ 35 S]GTP␥S loading was performed by incubating 25 g of membrane protein in TME buffer, 1 M GDP, and 30 nM [ 35 S]GTP␥S, plus agonist or vehicle, for 5 min at 30°C in a total volume of 100 l. Reactions were terminated by solubilizing the membranes for 30 min at 4°C in IP/Stop buffer (150 mM NaCl, 0.5% Nonidet P-40, 20 mM MgCl 2 , 100 M GDP, 100 M GTP, 1% aprotinin, 50 mM Tris-HCl, pH 7.5). Specific G protein subunits were isolated by immunoprecipitation for 1 h at 4°C using monoclonal antisera specific for G␣ s , G␣ q/11 , or G␣ i1/2 , collected on protein A-Sepharose. Immune complexes on Sepharose were washed three times with IP/Stop buffer, and [ 35 S]GTP␥S bound to the immunoprecipitated G␣ subunits was determined by liquid scintillation counting. [ 35 S]GTP␥S binding in the presence of 25 mM MgCl 2 was used as a positive control. Immunoprecipitations performed in the absence of primary antibody were used to determine nonspecific background.
cAMP Production-Appropriately transfected HEK-293 or COS-7 cells were split into 6-well plates and serum-starved overnight. Monolayers were preincubated with 1 M 3-isobutyl-1-methylxanthine for 15 min at 37°C, prior to stimulation with agonist for 6 -10 min as de-scribed in the figure legends. Reactions were terminated by aspirating medium and adding 250 l/well of cAMP buffer (4 mM EDTA, 50 mM Tris-HCl, pH 7.5) on ice. Monolayers were collected by scraping into Eppendorf tubes, boiled for 10 min, and clarified by microcentrifugation at 14,000 rpm for 15 min. The cAMP content of the supernatants was determined according to the manufacturer's instructions using the Biotrack [ 3 H]cAMP Assay System from Amersham Biosciences (22). Data were normalized to protein content as determined by Bradford assay of the cell lysates and expressed as pmol of cAMP/mg cell protein.
Phosphatidylinositol Hydrolysis-Appropriately transfected HEK-293 or COS-7 cells were split into 6-well plates and incubated for 18 -24 h with myo-[ 3 H]inositol at 4 Ci/ml in low serum growth medium. After labeling, cells were washed once with phosphate-buffered saline (PBS) and preincubated for 1 h in PBS at 37°C followed by fresh PBS containing 20 mM LiCl for 20 min. Cells were then stimulated for 1 h with agonist. Reactions were terminated by the addition of 1.0 ml of 0.4 M perchloric acid and neutralized with 0.4 ml of 0.72 M KOH and 0.6 M KHCO 3 . Total inositol phosphates were isolated by anion exchange chromatography on Dowex AG1-X8 columns and quantified by liquid scintillation spectroscopy, as described (11).
Phosphorylation of ERK1/2-Appropriately transfected COS-7 cells were split to 6-well plates and incubated for 18 -24 h in low serum growth medium in the presence or absence of inhibitors, as indicated. Agonist stimulation was carried out for 5 min, after which monolayers were washed once in 4°C PBS and lysed in 200 l of Laemmli sample buffer. For the determination of total cellular ERK1/2 and phospho-ERK1/2, aliquots containing ϳ20 g of cell protein were resolved by SDS-PAGE. ERK1/2 and phospho-ERK1/2 were detected by protein immunoblotting using polyclonal anti-ERK1/2 and anti-phospho-ERK1/2 antisera, respectively, with horseradish peroxidase-conjugated polyclonal donkey anti-rabbit IgG used as secondary antibody. Immune complexes were visualized by enzyme-linked chemiluminescence and quantified using a Fluor-S MultiImager. In each experiment, equal loading of ERK1/2 protein was confirmed by probing parallel immunoblots using anti-ERK1/2 antisera.

Cellular Expression of a Polypeptide Derived from the Carboxyl Terminus of G␣ s Inhibits G s -coupled Receptor Signaling by Blocking Receptor-G Protein
Coupling-To create a peptide inhibitor of receptor-G s coupling, we initially prepared a series of minigene constructs encoding 59, 83, and 110 amino acid polypeptides derived from the carboxyl terminus of bovine G␣ s . These polypeptides contain the major region of G␣ s thought to mediate contact with the intracellular domains of GPCRs but lack the sequences that contact adenylyl cyclases. As shown schematically in Fig. 1A, each minigene was composed of a minimal Kozak sequence, followed by the G␣ s -derived cDNA and a 3Ј-untranslated region. To facilitate detection of the expressed polypeptides, HA or GST epitopes were incorporated into the amino termini of each construct. Transient expression studies in COS-7 cells revealed robust expression of the 83-and 110-amino acid G␣ s -derived polypeptides. We were unable to detect expression of the 59-amino acid construct, suggesting that the polypeptide product was subject to rapid intracellular degradation. To determine whether expression of G␣ s -derived peptides affected signaling by a G s -coupled GPCR, we measured basal and agonist-stimulated cAMP production in COS-7 cells transiently expressing the D 1A dopamine receptor and either the GST-G␣ s -(313-395), HA-G␣ s -(313-395), or HA-G␣ s -(313-395) -CVLS minigenes. As shown in Fig. 1D, the GST-G␣ s -(313-395) peptide had no significant effect on D 1A receptor-mediated cAMP production, despite robust levels of expression. In contrast, expression of either HA epitope-tagged version of the construct led to a marked reduction in the cAMP response. Interestingly, addition of the prenylation sequence CVLS to the carboxyl terminus of the HA-G␣ s -(313-395) peptide, which might be expected to enhance membrane localization of the peptide, did not significantly increase its effectiveness. Based upon these data, we selected the unmodified HA epitope-tagged version of the 83-amino acid polypeptide G␣ s -(313-395) (GsCT) for further characterization.
The basic unit of heptahelical receptor signaling consists of receptor, heterotrimeric G protein, and effector. As depicted schematically for the ␤ 2 AR-G s -adenylyl cyclase module in 2A, it is possible to stimulate cAMP production in cells either by applying agonist, by activating G s directly using cholera toxin, or by activating adenylyl cyclase directly using forskolin.
To determine the effect of the GsCT polypeptide on adenylyl cyclase activation by ␤ 2 ARs, we employed a stable GsCT-expressing HEK-293 cell line. Fig. 2B compares the cAMP response of parental and GsCT-expressing HEK-293 cells to stimulation with the ␤ 2 AR agonist, isoproterenol, cholera toxin, or forskolin. In the presence of GsCT, isoproterenol-stimulated cAMP production stimulation was attenuated by ϳ68% compared with parental HEK-293 cells. In contrast, cAMP production occurring in response to receptor-independent activation of G s with cholera toxin, or of adenylyl cyclase with forskolin, was indistinguishable between the two cell lines. These data suggest that the GsCT peptide inhibits receptor-G protein coupling without directly impairing G protein or adenylyl cyclase function.
To elucidate further the mechanism of the inhibition produced by GsCT expression, we assayed the effect of the peptide on the affinity of ␤ 2 AR for agonist binding. As shown in Fig. 3A, the GsCT peptide, like the endogenous G␣ s protein, was present almost exclusively in the plasma membrane fraction following cell fractionation. Neither agonist exposure nor coexpression of FLAG epitope-tagged ␤ 2 AR increased the amount of GsCT in the membrane fraction, suggesting that the peptide inherently partitions into the membrane. As shown in Fig. 3B, the GsCT peptide specifically immunoprecipitates with the FLAG-␤ 2 AR from cotransfected cells, suggesting that once associated with the membrane, the peptide is capable of binding to the receptor.
In the absence of exogenous guanine nucleotide, many GPCRs exhibit characteristic high and low affinity states for agonist binding. The high affinity state is thought to represent pre-coupling of the GPCR to GDP-bound heterotrimeric G protein, whereas the low affinity state represents free GPCR. In the presence of a nonhydrolyzable GTP analogue, such as GTP␥S, which causes irreversible dissociation of G protein subunits, only the low affinity state of the receptor is present. Fig. 3C compares competition binding curves generated for the displacement of the ␤ 2 AR antagonist [ 125 I]-cyanopindolol by isoproterenol in COS-7 cells membranes in the presence of either GTP␥S or GsCT. In control membranes, the competition binding curve fits a two-site model with the high affinity site composing 22% of the total specific [ 125 I]-cyanopindolol-binding sites. In the presence of GTP␥S, the curve was shifted to the right, with only a single low affinity site present. In membranes from cells expressing the GsCT, the curve was similarly right-shifted, such that the high affinity site composed only 7% of the total. No significant differences were detected in the EC 50 values for the high and low affinity sites between control and GsCT-containing membranes. These data strongly support the hypothesis that GsCT binding to the ␤ 2 AR precludes receptor-G protein coupling.
Because GsCT expression appeared to target the receptor-G protein interface, we sought to determine whether the effect of GsCT expression was specific for G s by assaying receptor-stimulated [ 35 S]GTP␥S loading of endogenous G proteins in membranes isolated from parental and GsCT-expressing HEK-293 cells. For these assays, endogenous ␤ 2 ARs or transiently expressed ␣ 1B ARs and ␣ 2A ARs were employed to stimulate the endogenous pools of G s , G q/11 , and G i , respectively. As shown in Fig. 4A, membranes from GsCT-expressing cells showed a 72% decrement in the isoproterenol-induced increase in [ 35 S]GTP␥S loading of G␣ s compared with membranes from parental cells, with no significant effect on basal G␣ s loading. As shown in Fig.  4, B and C, no significant differences in basal or agoniststimulated G protein loading were observed when ␣ 1B AR-mediated G␣ q/11 and ␣ 2A AR-mediated G␣ i loading in GsCT expressing and parental cells were compared. Thus, GsCT expression led to G s -specific inhibition of receptor-G protein coupling.
GsCT Expression Results in G Protein-specific Inhibition of Heptahelical Receptor Signaling-If expression of the GsCT polypeptide selectively uncouples heptahelical receptors from G s , one would expect it to inhibit the generation of G s -dependent, but not G q/11 -or G i -dependent second messengers after stimulation of receptors coupled to these G protein pools. To test this hypothesis, we employed a transfected COS-7 cell system in which various heptahelical receptors were transiently expressed in the presence or absence of GsCT. Fig. 5A depicts the effects of increasing GsCT expression on cAMP production in response to stimulation of coexpressed G s -coupled D 1A dopamine receptors in COS-7 cells. At the highest levels of expression, GsCT inhibited D 1A receptor-mediated cAMP production to an extent comparable with that obtained by expression of a 59-amino acid polypeptide derived from the third intracellular domain of the D 1A receptor (D 1A R3i). We have shown previously that the D 1A R3i peptide, which represents the receptor side of the putative receptor-G protein interface, inhibits D 1A receptor-mediated cAMP production when expressed in HEK-293 and COS-7 cells (10, 11).  In the presence of either polypeptide, 6-chloro-PB hydrobromide-stimulated cAMP production was inhibited by at least 70% at each agonist concentration tested. The observed inhibition was not surmountable by even supersaturating concentrations of agonist. As shown in Fig. 5C, similar, apparently noncompetitive inhibition of ␤ 2 AR-mediated cAMP production was observed in the stable GsCT-expressing HEK-293 cell line. In the GsCT-expressing cells, isoproterenol-stimulated cAMP production was attenuated by at least 66% at each agonist concentration.
For the overexpressed D 1A R in COS-7 cells, and to a lesser FIG. 3. Effect of GsCT on high affinity agonist binding to the ␤ 2 AR. A, partitioning of GsCT between cytosol and membrane fractions. COS-7 cells were transfected with the HA-GsCT minigene (8 g/100-mm dish) in the presence or absence of FLAG-␤ 2 AR (2 g/dish) and treated for 5 min with isoproterenol (10 M) or vehicle prior to the preparation plasma membrane and cytosolic fractions. Immunoblots of 2% of the protein from each fraction were performed using antisera directed against the carboxyl terminus of G␣ s . The position of the endogenous p45 and p52 isoforms of G␣ s , as well as the GsCT minigene product, are as indicated. B, coprecipitation of HA-GsCT with FLAG-␤ 2 AR. COS-7 cells were transfected with the HA-GsCT minigene (8 g/100-mm dish) and FLAG-␤ 2 AR (2 g/dish), alone or in combination, as indicated. FLAG immunoprecipitates (IP) were subjected to immunoblotting (IB) using polyclonal anti-FLAG (left panel) and polyclonal anti-HA antisera (right panel) to detect the FLAG-␤ 2 AR and HA-GsCT peptide, respectively. The positions of FLAG-␤ 2 AR and coprecipitated HA-GsCT peptide are as indicated. C, agonist displacement curves for endogenous ␤ 2 AR in COS-7 cell membranes in the presence or absence of GTP␥S or HA-GsCT. Isoproterenol (Iso) displacement of [ 125 I]cyanopindolol ( 125 I-CYP) was performed using plasma membranes prepared from untransfected COS-7 cells or cells transfected with GsCT (20 g/150-mm dish). Agonist affinities and abundance were calculated by nonlinear regression analysis with one-and two-site models. Calculated values were as follows: control membranes, high affinity site 5. To determine whether this effect was due to inhibition of basal receptor-G s coupling or to an additional receptor-independent effect of GsCT, we compared the effect of GsCT expression with that of the ␤ 2 AR inverse agonist ICI118551 (23,24). In these assays we employed a constitutively activated point mutant of the ␤ 2 AR, ␤ 2 ARonco (23), because the higher basal levels of cAMP generated by the mutated receptor facilitated measurement of the effects of GsCT and ICI118551 on basal cAMP. As shown in Fig. 5D, basal cAMP production in COS-7 cells expressing the ␤ 2 ARonco was increased 4-fold in the presence of agonist and inhibited by 66% in the presence of maximally effective concentrations of ICI118551. As with the wild type ␤ 2 AR, both basal and agonist-stimulated cAMP production were attenuated in cells expressing GsCT. Treatment with ICI118551 had little additional effect on cAMP levels in the GsCT-expressing cells. The cAMP response to a submaximal dose of forskolin was equivalent between the two cell populations. The lack of additivity of the effects of GsCT and ICI118551 on ␤ 2 ARonco signaling suggests that the predominant effect of the GsCT is mediated through its effects on receptor-G protein coupling.
The apparently noncompetitive pattern of inhibition we ob-served is consistent with expression of an inhibitor that competes with the endogenous G protein pool for access to ligandbound receptor. Increasing agonist concentration would have no effect on the ratio of GsCT to functional G s heterotrimer and would thus not be expected to surmount the inhibitory effect of the polypeptide. Consistent with this, we found that the maximal extent of GsCT-induced inhibition of D 1A receptor-mediated cAMP production did vary with the level of receptor expression. At a D 1A receptor density of Ͻ0.75 pmol/mg, the cAMP response to saturating concentrations of agonist was almost completely blocked by high levels of GsCT expression. Increasing D 1A receptor expression to levels Ͼ1.5 pmol/mg partially overcame the inhibition (data not shown). In the presence of a saturating concentration of agonist, increasing receptor expression might increase the likelihood of an activated receptor encountering a functional G s heterotrimer, leading to less inhibition of second messenger generation. The ␣ 1B AR stimulates phosphatidylinositol (PI) hydrolysis primarily by G q/11 -dependent activation of the phospholipase C (PL-C) ␤1 isoform (25). As shown in Fig. 6A, ␣ 1B AR-mediated PI hydrolysis is unaffected by coexpression of increasing amounts of the GsCT polypeptide. The ␣ 2A AR weakly stimulates PI hydrolysis by G i -dependent activation of the PL-C ␤ 2 and ␤ 3 isoforms (26). As shown in Fig. 6B, ␣ 2A AR-mediated PI hydrolysis was likewise unaffected by GsCT expression. Collectively, these data suggest that expression of the GsCT polypeptide produces G protein-specific inhibition of heptahelical receptor-G protein coupling. Its effects are generalizable to multiple G s -coupled receptors, in that ␤ 2 AR and D 1A dopamine receptor cAMP production are similarly affected, but are specific for signals mediated by G s , in that G q/11 -and G i -dependent PI hydrolysis is unimpaired.
Use of GsCT to Examine the Contribution of G s to ERK Activation by ␤ 2 Adrenergic, ␣ 1B Adrenergic, and ␣ 2A Adrenergic Receptors-The role of G s proteins in GPCR-mediated ERK activation is complex. As depicted schematically in Fig. 7, previous studies have indicated that activation of protein kinase A (PKA) by G s -coupled receptors can produce both stimulation and inhibition of ERK activity. In HEK-293 cells (27), cardiac myocytes (28), and pancreatic acinar cells (29), ␤ 2 AR-mediated ERK activation involves both PKA and activation of pertussis toxin-sensitive G proteins. It has been proposed that phosphorylation of the ␤ 2 AR by PKA switches receptor coupling from G s to G i , allowing the receptor to mediate pertussis toxin-sensitive ERK1/2 activation through a G␤␥ subunit-dependent pathway (27). On the other hand, PKA-mediated phosphorylation of Raf-1 has been shown to attenuate growth factor-stimulated ERK activation in several cell types (30 -33). Thus, the net effect of G s stimulation on GPCR-mediated ERK activation likely reflects a balance between two opposing mechanisms of regulation.
Having determined that expression of the GsCT polypeptide leads to selective inhibition of G s -mediated signaling, we employed the construct to examine the contribution of G␣ s to ERK activation by ␤ 2 , ␣ 2A , and ␣ 1B adrenergic receptors. For the ␤ 2 AR, which is endogenously expressed, we compared isoproterenol-stimulated ERK1/2 phosphorylation in parental HEK-293 cells with that in stable GsCT-expressing HEK-293 cells. As shown in Fig. 8A, ␤ 2 AR-mediated ERK phosphorylation in HEK-293 cells was inhibited by pretreatment with either the PKA inhibitor, H89, or with pertussis toxin, consistent with the previously described roles of PKA and G i in the pathway (27). When isoproterenol-stimulated ERK1/2 phosphorylation was compared in parental and GsCT-expressing HEK-293 cells, a significant reduction was observed in the cells expressing the GsCT peptide. These data, shown in Fig. 8B, are consistent with the proposed requirement for G s activation in ␤ 2 AR signaling to ERK.
The ␣ 2A AR couples to G i/o family G proteins and in COS-7 cells mediates ERK activation through a pertussis toxin-sensitive pathway that is blocked by expression of a G␤␥ subunit sequestrant polypeptide derived from the carboxyl terminus of G protein-coupled receptor kinase 2 (34). However, the ␣ 2A AR also couples to G s , particularly at high levels of receptor expression (35). In contrast to the ␤ 2 AR, ␣ 2A AR coupling to both G i and G s is a constitutive property of the receptor, not one that is modulated by PKA phosphorylation. Furthermore, G s activation apparently antagonizes GPCR-stimulated ERK activation in COS-7 cells, because expression of an activated mutant of G␣ s , or treatment with the cell-permeant cAMP analog, 8-bromo-cAMP, attenuates ERK activation in response to either isoproterenol or epidermal growth factor in this system (33). As shown in Fig. 9A, ␣ 2A ARs transiently expressed in COS-7 cells, like ␤ 2 ARs in HEK-293 cells, activate ERK1/2 via pertussis toxin-sensitive G proteins. However, in contrast to the ␤ 2 AR system, treatment with H89 enhances, rather than inhibits, ␣ 2A AR-mediated ERK activation. This presumably reflects relief of the inhibitory effect of PKA on ERK activation that results from phosphorylation of c-Raf1. As shown in Fig.  9B, transfection of COS-7 cells with increasing amounts of the GsCT plasmid, like H89 treatment, caused a progressive enhancement of ␣ 2A AR-mediated ERK phosphorylation. In COS-7 cells, stimulation of transiently expressed ␣ 1B ARs leads to ERK activation that is pertussis toxin-insensitive (36) but blocked by expression of a polypeptide derived from the carboxyl-terminal 55 amino acids of G␣ q (18). As shown in Fig.  9C, transfection of COS-7 cells with increasing amounts of the GsCT plasmid had no effect on ␣ 1B AR-mediated ERK phosphorylation. Thus, ERK1/2 activation by a receptor that does not activate G s was unaffected by GsCT expression. DISCUSSION The precise structural determinants underlying activation of heterotrimeric G proteins by heptahelical receptors are incompletely understood. Crystallographic analysis of the structure of G␣ s has indicated that receptor coupling specificity is likely determined by a surface formed by the continuous carboxylterminal ␣-helix between Asp-368 and Leu-394, and the loop between the ␣5-helix and ␤6-strand (6). Contact between the G␣ subunit and the second and third intracellular domains of heptahelical receptors determines the efficiency and specificity of the receptor-G protein interaction (3,37,38). The G␣ subunit carboxyl-terminal helix may insert into a cavity between the third and sixth receptor transmembrane domains of the heptahelical receptor bundle that forms as a consequence of agonist-induced conformational changes (39). NMR studies have demonstrated that short polypeptides derived from the G␣ s carboxyl terminus form stable ␣-helices in solution. In isolated plasma membranes, 11-amino acid peptides representing the carboxyl termini of G␣ i1/2 or G␣ o modulate ligand binding to the adenosine A 1 receptor by disrupting the high affinity receptor-G protein complex (16). Similarly, modified 16 -21-amino acid peptides derived from the carboxyl terminus of G␣ s inhibit high affinity agonist binding to the adenosine A 2A receptors, and impair A 2A receptor-mediated adenylyl cyclase activation (17). These data suggest that the isolated carboxyl-terminal ␣-helix can interact with a receptor in a manner that precludes productive receptor-G protein coupling.
We have examined the mechanism of action and functional consequences of expression of an 83-amino acid polypeptide derived from the carboxyl terminus of G␣ s in intact cells. Expression of the GsCT peptide impaired adenylyl cyclase activation by G s -coupled ␤ 2 adrenergic and D 1A dopamine receptors, without affecting the response to cholera toxin or forskolin, suggesting that the peptide specifically impairs receptor-G protein coupling. At a constant level of receptor expression, the inhibition was not surmountable by increasing agonist concentration. The magnitude of the effect was partially reversed by increasing receptor density, consistent with the hypothesis that the peptide competes with the endogenous G s pool for access to ligand-occupied receptors. Furthermore, the inhibition was apparently specific for G s , because PI hydrolysis induced by stimulation of the G q/11 -coupled ␣ 1B AR and G i -coupled ␣ 2A AR was unaffected by GsCT expression.
A significant aspect of the approach of using receptor-or G protein-derived peptides to inhibit heptahelical receptor signaling is that the resulting antagonism affects a class of G protein, rather than a specific receptor. Such reagents differ from pharmacologic antagonists of ligand binding in that expression of a single polypeptide should be able to uncouple multiple receptors from a single G protein pool. In this regard, their function is more like B. pertussis toxin, which uncouples all G i/o family proteins from their cognate receptors by catalyzing the ADPribosylation of a carboxyl-terminal cysteine residue on the G␣ subunit.
Because of their ability to selectively uncouple specific G proteins from multiple receptors, peptide inhibitors of receptor-G protein coupling may be useful for determining the contribution of a given G protein pool to signaling by a receptor that couples to multiple G proteins. Minigene constructs encoding the carboxyl termini of G␣ i , G␣ q , G␣ 12 , and G␣ 13 have recently been employed to examine the contribution of different G protein pools to second messenger generation by the thrombin receptor in endothelial cells (20). We have employed GsCT expression to examine the role of G␣ s in a complex process, activation of the ERK MAP kinase cascade in fibroblasts, where G s activation has been reported previously to produce both stimulation and inhibition of ERK activity. Consistent with previous reports, we found that ␤ 2 AR-mediated ERK activation, which is blocked by PKA inhibition, is inhibited by GsCT, whereas ␣ 2A AR-mediated ERK activation, which is accentuated by PKA inhibition, is enhanced in cells expressing GsCT (27)(28)(29)(30)(31)(32)(33). These data support a dual role for PKA in ERK activation by the ␤ 2 AR, where PKA phosphorylation of the receptor promotes receptor-G i coupling and pertussis toxinsensitive ERK activation (27), but where PKA phosphorylation of Raf1 attenuates ERK activation downstream of Ras (30 -33). In the case of the constitutively G i /G s -coupled ␣ 2A AR, only the downstream inhibitory effect of G s activation, which is relieved by GsCT expression, is discernible.
Tissue-specific expression of peptide G protein inhibitors has already provided valuable information about the roles of individual G proteins in complex physiologic responses in vivo. Cardiomyocyte-specific expression of a 55-amino acid peptide derived from the carboxyl terminus of G␣ q reduces cardiac hypertrophy (18) and inhibits activation of the ERK and c-Jun amino-terminal kinase MAP kinase cascades (40), in response to surgically induced pressure overload in a transgenic murine model, underscoring the important role of G q/11 proteins in this process. Recombinant adenovirus-mediated expression of a G␤␥ subunit sequestrant polypeptide derived from the carboxyl terminus of G protein-coupled receptor kinase 2 (41), which results in generic inhibition of G␤␥ subunit-mediated signaling events (34,42), blocks ERK activation and vascular smooth muscle hypertrophy in a rat carotid artery model of vascular restenosis (43). The availability of specific polypeptide inhibitors of G␣ s signaling, such as the GsCT minigene, may provide the opportunity to obtain similar insights into the complex roles of G s in control of cellular hypertrophy and proliferation in a variety of tissues.